TEST FACILITY TO INVESTIGATE PLUME-REGOLITH INTERACTIONS
Craig White1, Hossein Zare-Behtash1, Konstantinos Kontis1, Takahiro Ukai2, Jim Merrifield3, David Evans3, Ian Coxhill4,
Tobias Langener5, Jeroen Van den Eynde5
1 School of Engineering, University of Glasgow, Glasgow, UK
2 Department of Mechanical Engineering, Osaka Institute of Technology, Osaka, Japan
3 Fluid Gravity Engineering Ltd, West Street, Emsworth, Hampshire PO10 7DX, UK
4 Nammo Westcott, Westcott Venture Park, Westcott, Buckinghamshire, UK
5 ESA-ESTEC, Keplerlaan 1, 2201 AZ Noordwijk, Netherlands
For both sample return and insitu-analysis missions it is
vitally important to understand the physical phenomena
related to the propulsion subsystem and terrain interactions
during landing of the spacecraft. The plumes of the retro-
rockets will cause erosion and chemical contamination to
the celestial body’s surface which will affect the landing
manoeuvres and measurements of the regolith in the vicinity
of the landing site. Furthermore, the generated regolith dust
clouds and the reflected plume flow can affect the spacecraft
force and torque balance as well as posing problems for the
proximity navigation equipment and sensitive external
surfaces. To investigate the interaction between rocket
engine plume and regolith at a fundamental level, in support
to all planetary and lunar landing environments, such as the
Phobos Sample Return mission, a dedicated facility is
designed, manufactured, and housed at the University of
The results presented will cover the design calculations
of the vacuum facility with the added mass flow from
pulsating hypersonic nozzles. The methodology behind the
specially designed heat exchanger operating at different
temperatures, pressures, and frequencies representative of
the various landing scenarios is also presented. Analysis to
determine the most suitable material to act as regolith
simulant, due to the variations in Earth’s specific gravity
compared to Martian and airless bodies, is also discussed.
Index Terms— Plume-regolith interaction, high-
vacuum facility, Regolith.
During sample return or insitu-analysis missions, the plumes
of the retro-rockets will cause erosion and chemical
contamination to the celestial body’s surface which will
affect the landing manoeuvres and measurements of the
regolith in the vicinity of the landing site. Due to the
chemical composition of the plume gases, contamination of
the soil needs to be minimised. Furthermore, the generated
regolith dust clouds and the reflected plume flow can affect
the spacecraft force and torque balance as well as posing
problems for the proximity navigation equipment and
sensitive external surfaces.
The objective of this activity is to investigate the
interaction between rocket engine plume and regolith at a
fundamental level, in support to all planetary and lunar
landing environments such as the Phobos Sample Return
mission. These studies are vitally important as the
interaction between the hovering and landing plumes with
the regolith can have a severe impact on the mission
objectives and also the engine performance. To achieve
these goals, a dedicated facility, with a total volume of
approximately 72 m3, is designed, manufactured, and
housed at the University of Glasgow. The volume is split
between the test chamber, a 12 m3 axisymmetric chamber
with access ports, and a 60 m3 buffer tank. The purpose of
the buffer tank is to keep the pressure rise in the test
chamber, during the addition of any mass flow, within an
The facility is designed with two missions in mind. The
first case is representative of landing on an airless body, and
the second scenario is landing on Mars. Hence, two different
mass flow cases which are scaled versions of the proposed
thrusters are analysed: a 1 g/s mass flow rate, with a nozzle
exit Mach number of 6.6 for landing on an airless body, and
18.9 g/s with an exit Mach number of 5.84 for landing on
Mars. Previous studies of a similar nature have never
attempted to match the nozzle exit Reynolds numbers, so in
this work the gas is heated in the stagnation chamber to
achieve Reynolds number similtude. The gas passing
through the nozzle must be heated to 700 K and 1000 K for
the Martian and airless body cases, respectively, to satisfy
the Reynolds number matching. Strouhal similtude is also
important as the motors will operate in a pulsed mode and
this means that the heater assembly must also be low
volume in order to be able to pulse the scaled nozzles at the
The aforementioned requirements led to the ultimate
design of the high-vacuum large volume facility shown
schematically in Figure 1. The system is fabricated from
stainless steel to minimise the effects of outgassing and
ensure longevity of the setup.
Figure 1. Schematic of plume-regolith facility showing test
chamber, buffer tank, pumping system, and conductance pipe.
2. AIRLESS BODY SCENARIO
For the airless body test case, Nitrogen gas is added into the
test chamber at a rate of 1 g/s for 2.5 seconds. The pressure
inside the test chamber must always maintain a pressure
ratio between the total pressure at the nozzle throat and the
background pressure of more than 100,000. To assists with
maintaining the pressure a buffer tank with an additional
volume of 60 m3 is added to the system. The layout of the
test chamber and buffer tank is dictated by the space
available on site. The design of the buffer tank and its
connection to the test chamber, as shown in Figure 2, takes
into account the conductance in the pipework.
Figure 2. Mass flow from test chamber to buffer tank.
Since the nozzle is fired within the test chamber, at the
onset of Mode 1 (airless body) operation the pressure in the
test chamber will rise ahead of that at the inlet of the Roots
pump or in the buffer tank. As the test chamber pressure
rises however, the pressure differential across the Mode 1
conductance drives a flow over the Roots pump and into the
buffer tank. If the conductance is split in to two
components; C1 between test chamber and Roots inlet and
C2 between Roots inlet and buffer tank; then the flow from
the test chamber to the Roots inlet is given by
the net flow into the test chamber is therefore:
where 87 is 1g/s of Nitrogen expressed in SI units, Pa m3/s,
and the rate of pressure rise in the test chamber is:
Whilst this may look like a soluble first order
differential equation, it should be remembered that the
Roots inlet pressure is also varying with time and is
dependent on the volumetric capacity characteristic of the
Roots pump, for which the exact mathematical form is not
available to the author. In practice solutions implemented
were discrete element approximations only.
As the Roots pump starts close to its ultimate pressure
and has effectively no pumping speed at this pressure (all
the work done is in compressing exhaust gases to maintain
the pressure differential across the pump), this flow will
cause a rise in the Roots pump inlet pressure. This in turn
will generate a pressure differential between the Roots pump
inlet and the buffer tank, and thereby a flow from the Roots
pump inlet to the buffer tank of:
so that there will be a rise in buffer tank pressure of:
Through calculation of the pumping system behaviour
as a function of inlet pressure, provided by BUSCH, the
effective pumping speed of the pumping group, and
therefore the net flow into the Roots pump as the pressure
rises is give by:
substituting Eq. 1 and Eq. 4 into Eq. 6, this can be
rearranged to give the Roots pressure a function of the test
and buffer tank pressures:
this equation is linear, so the same relation holds in the time
differential of the equation.
%&'% − $
!"#$" =87 − )
"#$" − )
87 − "
#$%# − "
%&&'( − $
"',,-. / "#$%%&' 01
!"− !$= &
"##$% * ()* (.
We now have time dependent pressure rise formulae for
the test chamber, Roots pump inlet and buffer tank in the
presence of a 1 g/s Nitrogen gas feed into the test chamber.
These can be applied to predict the performance of the
system in airless body mode. The predicted pressure rise is
graphically illustrated in Figure 3. As evident from the
figure, the system is capable of maintaining the desired
pressure of 4 Pa (0.04 mbar) within the test chamber for the
duration of the experiments: 2.5 seconds.
Figure 4 displays the measured pressure rise within the
test chamber during a 1 g/s addition of Nitrogen gas. The
starting background pressure within the test chamber was
recorded at 0.9 Pa. The initial rapid rise in pressure is
attributed to the high-speed propagation of the Nitrogen
molecules exiting the nozzle and impinging on the surface
of the pressure sensor mounted on the top of the test section.
Figure 3. Theoretical pressure rise for airless body scenario.
Comparison of the final pressure after 2.5seconds of
Nitrogen addition between the measured and calculated
values yields an acceptable agreement between the two
Figure 4. Measured pressure rise for airless body scenario, two
3. MARTIAN SCENARIO
For simulation of the Martian environment, a background
pressure of 600 Pa (6 mbar) must be maintained whilst
Nitrogen gas is added at a much higher mass flow rate of
18.9 g/s equivalent to 1644 Pa m3/s. To this end, a pressure
control valve, identified in Figure 5, is installed at the
exhaust of the test chamber and is controlled by a
capacitance manometer metering the opening of the valve to
maintain the desired test chamber pressure.
Figure 5. Pressure control case for Martian scenario.
There are, in effect, three phases of operation with the
Martian mode of operation:
• Closed: initially the vacuum level on the test chamber
is high (low pressure) and the control valve is closed to
allow the pressure within the test chamber to rise to
• Control: once a chamber pressure of 600 Pa is reached,
the control valve begins to open and its rate of opening
is metered to maintain a pressure of 600 Pa in the test
• Open: when the valve is fully open and there is no
longer a sufficient pressure difference to maintain an
18.9g/s Nitrogen flow through the conductance, the
pressure in both the test chamber and ballast chamber
begin to rise but, due to the large overall volume, the
rise is slow.
3.1. Closed phase
The control valve selected does not fully close but rather,
has a minimum conductance of 4 l/s. Given that the test
chamber pressure must, by definition, be less than 600 Pa
during this phase, the maximum flow through the valve is
therefore 2400 Pa l/s, or 1.44 SLM. At this flow the inlet of
the Roots booster will be close to ultimate pressure and can
be neglected when calculating the flow from the chamber.
This flow is also only around 0.15% of the flow into the
chamber so can also be neglected in terms of its effect in
reducing the rate of pressure rise in the chamber. The
pressure rise in the test chamber during the closed phase is
a constant rate of approximately (130 Pa/s) (1.3 mbar/s).
3.2. Control phase
Once the test chamber pressure has reached 600 Pa, the
control valve starts to open. The Roots booster is connected
to the test chamber by a smaller conductance (DN 160) than
in the airless body case, so the Roots pump and buffer tank
remain at similar pressures, with a pressure drop at the
control valve and along the length of the conductance.
During this phase the flow out of the test chamber must be
equal to the flow into the test chamber in order to maintain a
constant pressure. The control valve therefore must meter
the conductance of the combination of the conductance pipe
and itself to:
with QMartian= 1644 Pam3/s and Ptest= 600 Pa. The pressure
rise within the ballast chamber and airless body conductance
pipework is therefore:
3.3. Open phase
As the conductance between the test chamber and Roots
pump is finite, after a further period of time the pressure
difference between the test chamber and Roots pump will
cease to be sufficient to sustain a flow of 18.9 g/s of
Nitrogen gas and the pressure in the test chamber will begin
to rise. Since conductance rises with pressure, the
conductance of the pipework in the open phase is now
sufficiently high such that a good approximation can be
obtained by calculating the average rate of pressure rise
throughout the system, and then calculating the pressure
differential between the two chambers. The net flow into the
However, again the pressures within the system are
sufficiently similar that the average system pressure is a
sufficient approximation for the Roots booster inlet
pressure, especially as the Roots booster inlet, buffer tank
and airless body conductance pipework make up
approximately 85% of the total system volume. The average
rate of pressure rise is therefore:
Assuming the test and buffer tank pressures rise
approximately in tandem, the net flow into the test chamber
leaving a net flow through the conductance of:
providing the values necessary to find the test chamber
pressure from the buffer tank pressure (approximating to the
average pressure) via the formula:
Figure 6 shows the Martian scenario simulation
obtained by applying Equations 8, 10, 12 and 15. The test
chamber reaches 600 Pa after 4.7 seconds and the control
phase – continuous operation at a stable 6 mbar – lasts a
further 26 seconds. As the control valve reaches its fully
open position, the pressure at the inlet of the Roots pump is
590 Pa (5.9 mbar) and continues to converge with the test
chamber pressure, validating the assumption that the test
and buffer tank pressures are very similar during the open
Figure 6. Theoretical pressure rise for Martian scenario.
Figure 7 displays the measures pressures inside the test
chamber, buffer tank, and the measured mass flow rate into
the test chamber. The green line in the graph shows the
operation of the control valve. At 10 volts the control valve
is fully open. As visible from the figure, the test chamber
and buffer tank begin at an initial pressure of 0.9 Pa. As
Nitrogen gas enters the test chamber, purple line of mass
flow rate, the pressure inside the test chamber begins to
!"#$%&#' = )"#$%&#' *%+,% − *$..%,
!"#$ = 1644 − *
()** + , "
!"#$%&"'($") *+,-- ./0
!"#$" % !
increase rapidly. When the pressure reaches 600 Pa the
valve begins to open (green line) to allow Nitrogen to flow
into the buffer tank and tries to maintain 600 Pa inside the
test chamber. However, due to the high mass flow rate,
starting at 20g/s and reducing to 19g/s over a 15 second
period, the target pressure of 600 Pa cannot be maintained in
the test chamber. The grey line shows the rising pressure
inside the buffer tank calculated from the measured voltage
(yellow line) of the vacuum sensor on the buffer tank. The
red line is the trigger signal.
Figure 7. Measured pressure rise for Martian body scenario,
20g/s N2 gas.
Figure 8 shows a repeat test at a reduced mass flow rate
of Nitrogen, starting from 17.2g/s and reducing to 16.4g/s
over 25 seconds. In the scenario, even though the starting
pressure within the test chamber is approximately 550 Pa
(this run was performed after the results of Figure 7), as
evident by the pressure inside the test chamber, the target
pressure of 600Pa is maintained for 15 seconds..
Figure 8. Measured pressure rise for Martian body scenario,
Due to the long connecting pipework between the
Nitrogen gas cylinder and the nozzle inside the test
chamber, there are pressure losses at play which make it
difficult to reach and maintain a constant mass flow into the
chamber. The mass flow rate also appears to be influenced
by the rising back pressure within the test chamber.
4. REGOLITH SIMULANTS
It is desired to conduct experiments on a simulated lunar,
Martian or Phobos regolith within the vacuum chamber. In
contrast to the bulk of experiments involving extra-
terrestrial dust particles, exploring their impact on the
flowfields encountered during planetary entry (gravitational
forces being insignificant in these environments), the
simulation of behaviours on the surface will require a
realistic representation of the gravitational field. Since this
is substantially weaker on the extra-terrestrial bodies under
consideration, the density must be reduced significantly for
the weight to match. The factors of density to be applied in
each environment are 0.165, 0.387, and 0.000591 for lunar,
Martian, and Phobos environments, respectively.
4.1. Lunar regolith
Graf  published a handbook of lunar soil grain size data
taken from the Apollo and Luna missions. 287 analyses of
143 separate samples were presented. Graphic measures
were used to obtain statistical data, which was presented for
each sample. Considerable variation existed, but modal
values around the 60 microns radius mark were typical.
Nakashima et al  carried out experiments on the cutting
resistance of lunar regolith under reduced gravity
conditions. This enabled a simulant of representative mass,
FJS-1, to be used. The particle size distribution lies in the
range of Apollo measurements, with a median value of
approximately 60 microns. In terms of the regolith
behaviour as the gravity was reduced, a linear relationship
between specific cutting resistance and gravity was
discovered, suggesting that merely reducing the density of a
regolith simulant will not adequately represent the effects of
For lunar regolith, the low factored density (0.511gcm3)
inhibits the use of traditional regolith simulants (JSC-1 etc.).
Glass microspheres manufactured by 3M  have been
chosen for the current work. The closest approaching the
requirements is the K46 grade, with a density of 0.46 g/cm3
and a median particle size of 40 microns.
4.2. Martian regolith
Greeley et al.  carried out an experimental investigation
of wind threshold speeds in a wind tunnel set up to replicate
the Martian environment as closely as possible. The effect
of reduced gravity was addressed by the use of crushed
walnut shells, which were found to be the most appropriate
of a number of candidate materials. Particle sizes of between
35 and 600 microns were tested. Crushed walnut shells, of
sizes between 61 and 88 microns, were also used by Greeley
and Leach  in an investigation of dust injection into the
Martian atmosphere. Mehta et al  also used crushed
walnut shells in their experimental investigation of the
‘explosive erosion’ process encountered during the landing
of the Phoenix spacecraft. Four different size distributions
were employed: poorly sorted fine sand (~160 microns
diameter), fine silt/dust (<15 microns diameter), large coarse
sand/dust (850-2500 microns diameter), and a bimodal
mixture of equal volumes of fine sand and fine silt/dust.
Crushed walnut shells will be used for the Martian
simulant, with a mesh No. 100, giving an average particle
size of 149 microns, which is in the range of particle
diameters outlined above.
4.3. Phobos regolith
No direct measurements of the Phobos regolith exist.
However, Gundlach and Blum  predict a value of 1.1
+0.9/-0.7 microns for the regolith, which is very small.
Other studies suggest a representative size of 10-100
The K1 series of 3M glass bubbles have been chosen for
their small density (around 0.125 g/cm3).
5. FUTURE WORK
Future work is concentrated on achieving the target mass
flow rate into the test chamber for the Martian scenario and
maintaining 600Pa throughout the duration of the tests.
Upon achieving this goal work will begin on development
of in-house heat exchanger to heat the Nitrogen gas to the
desired temperatures for the airless body and Martian cases
for Reynolds number similarity. Investigation of the plume-
regolith interaction is the final stage of this campaign.
The authors are grateful for the technical input and expertise
of BUSCH UK who worked tirelessly on this project,
especially Andrew Pearce, Christine Hewitt, Mortimer
Brown, and David Black. We are also indebted to the efforts
of Colin Roberts and Sriram Rengarajan who have
contributed greatly to this project.
 J.C. Graph, “Lunar soils grain size catalog,” NASA RP-1263,
 H. Nakashima, Y. Shioji, K, Tateyama, S. Aoki, H. Kanamort,
and T. Yokoyama, “Specific cutting resistance of lunar regolith
simulant under low gravity condition,” Journal of Space
Engineering 1, 2008.
 3M Energy and Advanced Materials Division.
 R. Greeley, R. Leach, B. White, J. Iversen, and J. Pollack,
“Threshold windspeeds for sand on Mars wind tunnel simulations,”
Geophysical Research Letters 7, 1980.
 R. Greeley, and R. Leach, “Steam injection of dust on Mars:
Laboratory simulations in reports of planetary geology program,”
NASA TM-80339, 1979.
 M. Mehta, N.O. Renno, J. Marshall, M.R. Grover, A. Sengupta,
N.A. Rusche, J.F. Kok, R.E. Arvidson, W.J. Markiewics, M.T.
Lemmon, and P. Smith, “Explosive erosion during the Phoenix
landing exposes subsurface water on Mars,” Icarus 211, 2011.
 B. Gundlach, and J. Blum, “A new method to determine the
grain size of planetary regolith,” Icarus 223, 2013.
 R.O. Kuzmin, T.V. Shingareva, and E.V. Zabalueva, “An
engineering model for the Phobos surface,” Solar System Research
 E. Kührt, B. Giese, H.U. Keller, and L.V. Ksanfomality,
“Interpretation of the KRFM-infrared measurements of Phobos,”
Icarus 96, 1992.